248 |
\subsection{Texture perception} |
\subsection{Texture perception} |
249 |
|
|
250 |
Psychological studies on texture perception have mostly concentrated |
Psychological studies on texture perception have mostly concentrated |
251 |
on texture discrimination, the ability of human observers to discriminate |
on \emph{texture discrimination}, the ability of human observers to |
252 |
pairs of textures. |
discriminate pairs of textures. |
253 |
% XXX: segregation vs. discrimination |
The term is often used interchangably with \emph{texture segregation}, |
254 |
|
the more specific task of finding the border between areas of |
255 |
First experiments on computer-generated, unnatural textures |
different textures (different phases of local characteristics at the |
256 |
in the 60s \cite{julesz62visualpattern} led to |
border can segregate otherwise indiscriminable textures). |
257 |
proposals of discrimination models based on |
|
258 |
$N$th order statistics of pixels and connectivity |
First experiments on computer-generated, unnatural textures in the 60s |
259 |
structures of certain micropatterns. |
\cite{julesz62visualpattern} led to proposals of discrimination models |
260 |
|
based on $N$th-order statistics (the joint distributions of all |
261 |
|
$N$-tuples of pixels for given $N$) and connectivity structures of |
262 |
|
certain micropatterns. |
263 |
|
|
264 |
Statistical modeling of textures as samples from a probability |
Statistical modeling of textures as samples from a probability |
265 |
distribution on a random field as already seen in \cite{julesz62visualpattern} |
distribution on a random field as already seen in \cite{julesz62visualpattern} |
266 |
in a simple form. |
in a simple form. |
267 |
The most popualar computational approach is Markov random fields |
The most popualar computational approach is Markov random fields |
268 |
\cite{cross83markov, geman84stochastic}, where a texture |
\cite{cross83markov, geman84stochastic}, where the value of each pixel |
269 |
is characterized by its local statistics. |
depends only on the values of its neighborhood (local characteristics). |
270 |
XXX: resolution-dependency? |
XXX: resolution-dependency? |
271 |
|
|
272 |
Attempt to explain texture perception by the densities of textons |
Attempt to explain texture perception by the densities of textons |
283 |
Essentially a bank of linear filters is applied to the texture followed |
Essentially a bank of linear filters is applied to the texture followed |
284 |
by a nonlinearity and then another set of filters. |
by a nonlinearity and then another set of filters. |
285 |
|
|
286 |
|
Mapping texture appearance to an Euclidian texture space |
287 |
|
(see \cite{gurnsey01texturespace} and the references therein): |
288 |
|
in the reported experiments, three dimensions have been sufficient |
289 |
|
to explain most of the variation in the similarity judgements for |
290 |
|
artificial textures. |
291 |
|
However, the texture stimuli have been somewhat simple |
292 |
|
(no color, lack of frequency-band interaction, etc.). |
293 |
|
For some natural texture sets, three dimensions have also been |
294 |
|
sufficient, but often the semantic connections cause the |
295 |
|
similarity to be context-dependant, making it hard to assess the |
296 |
|
dimensionality. |
297 |
|
% XXX: this is something we should experiment with our textures |
298 |
|
|
299 |
XXX: reviews |
XXX: reviews |
300 |
|
|
301 |
XXX: physiological knowledge of visual perception |
XXX: physiological knowledge of visual perception |